Joined: 01 Aug 2006
Location: Richland, Washington State, USA
|Posted: Wed Jul 13, 2011 12:59 am Post subject: Reflections of hard and soft light in a spider's eye
|This is a follow-up to Reflections of reality (in the eyes of a jumping spider). My purpose here is to illustrate and explain some of the reasons why "hard" light (not diffused) produces strange effects in stacks.
To quickly summarize, here is the eye of a spider shot with two very different lighting setups. In each row, there are two source images and the stacked result produced by Zerene Stacker PMax.
In the top row, the stacked result is clean -- it looks just like what you'd expect by "picking out the sharp bits" of each of the input files.
But the bottom row is very different. The stacked result is an ugly mess. Somehow the surface detail of the eye and the direct reflection of the flash head have both gotten wiped out, replaced with some sort of weird radial artifact.
What the heck is going on??
Some insight can be gained by looking at more of the individual frames.
First, let's look at a bunch of frames from the clean stack.
I think there are no surprises here. The eye and the reflection start off fuzzy as focus is far in front. Then gradually the surface of the eye comes into focus, and somewhat later the diffuser and objective. The whole process is very smooth and gradual.
Now let's look at a bunch of frames from the ugly mess stack.
Oh, that's what's going on! In every frame, the out-of-focus reflection of the bare flash has a hard edge. Even more interesting, the surface texture of the eye has reflected the flash as a blotchy pattern of light across the aperture of the lens, and following the usual rules for out-of-focus highlights, that blotchy pattern at the aperture is reproduced in the "blur circle" of the OOF flash reflection. These two processes combined cause a pattern of completely unreal "detail" to march across the image, and PMax dutifully collects it all up to produce the huge highlight with radial streaks that eventually appears in the result. To quote somebody or other, "It's not a pretty picture."
There's no new moral here. We already know that the cleanest stacks come from diffuse light, and failing to diffuse is asking for trouble. But perhaps this illustration adds a bit of detail to help explain why that's the case.
What we're seeing here is essentially a matter of bokeh. Using diffuse "soft" light, the large direct reflections go in and out of focus as smooth blurs that stacking software can easily distinguish from actual detail. Using very directional "hard" light, the tiny direct reflections go in and out of focus as moving patterns that look a lot like detail to be preserved.